Diagnosis, treatment and survival from bladder, upper urinary tract, and urethral cancers: real‐world findings from NHS England between 2013 and 2019

Objective We report NHS England data for patients with bladder cancer (BC), upper tract urothelial cancer (UTUC: renal pelvic and ureteric), and urethral cancers from 2013 to 2019. Materials and Methods Hospital episode statistics, waiting times, and cancer registrations were extracted from NHS Digital. Results Registrations included 128 823 individuals with BC, 16 018 with UTUC, and 2533 with urethral cancer. In 2019, 150 816 persons were living with a diagnosis of BC, of whom 113 067 (75.0%) were men, 85 117 (56.5%) were aged >75 years, and 95 553 (91.7%) were Caucasian. Incidence rates were stable (32.7–34.3 for BC, 3.9–4.2 for UTUC and 0.6–0.7 for urethral cancer per 100 000 population). Most patients 52 097 (mean [range] 41.3% [40.7–42.0%]) were referred outside the 2‐week‐wait pathway and 15 340 (mean [range] 12.2% [11.7–12.6%]) presented as emergencies. Surgery, radiotherapy, chemotherapy, or multimodal treatment use varied with disease stage, patient factors and Cancer Alliance. Between 27% and 29% (n = 6616) of muscle‐invasive BCs did not receive radical treatment. Survival rates reflected stage, grade, location, and tumour histology. Overall survival rates did not improve over time (relative change: 0.97, 95% confidence interval 0.97–0.97) at 2 years in contrast to other cancers. Conclusion The diagnostic pathway for BC needs improvement. Increases in survival might be delivered through greater use of radical treatment. NHS Digital data offers a population‐wide picture of this disease but does not allow individual outcomes to be matched with disease or patient features and key parameters can be missing or incomplete.


Background
Bladder cancers (BCs) and upper tract urinary cancers are common and important public health problems [1]. More than 550 000 new cases are diagnosed yearly worldwide [2]. These tumours are some of the most expensive human cancers to manage [3], and affected individuals report poor experiences [4,5] with reduced long-term quality of life [6]. BCs and upper tract urothelial cancers (UTUCs, including ureteric and renal pelvic cancers) often arise following exposure to smoking [7], occupational carcinogens [8], or diesel fumes. In addition, UTUCs may be part of Lynch syndrome or develop after exposure to aristolochic acids [9]. Most BCs, ureteric and renal pelvic cancers, and urethral cancers are urothelial cell carcinoma (UCC) in histological subtype and best stratified into non-muscle-invasive (NMI) and muscle-invasive (MI) cancers. The natural history of NMI and MI cancers differs widely, so treatments reflect risks of progression and metastases [9][10][11]. Non-urothelial bladder and upper tract tumours are often MI with poor outcomes [12].
Population-wide healthcare data have the potential to offer an objective understanding of the real-world presentation, treatment, and outcomes. Improvements in connectivity and analytical capacity, combined with lower costs, mean realworld data are becoming more common and available for interrogation [13]. Proponents suggest real-world data can replicate prospective randomised trials at low cost and with fewer ethical barriers [14]. As such, they are replacing custom-built, service-specific, audit-focused datasets that are limited in remit and reliant on enthusiasts [15]. However, completion rates may be problematic (missing data), outcomes missing (or misclassified), and patterns of care/ outcomes prone to bias. Here, we analyse the most contemporary dataset reflecting BC, UTUC and urethral cancer within NHS England. We hypothesise this will reveal trends in the treatment of these cancers and areas where improvement is needed and possible.

Data Extraction
We extracted data from the National Cancer Registration and Analysis Service (NCRAS) Data Repository, NHS Digital (Appendix S1). We selected cancers, regardless of stage, grade and histology, of the renal pelvis (International Classification of Diseases, 10th edition    -4]). The route to diagnosis was presented for each cancer and allocated to one of: GP referral, 2-week-wait (2ww) pathway (defined as urgent GP referrals with a suspicion of cancer), emergency presentation, outpatients (other), screen detected, inpatient elective (where no earlier information can be found prior to admission from a waiting list, booked or planned), diagnosis by death certificate only, and unknown [16]

Analysis and Statistics
Data were extracted and analysed within Excel for Mac (Version 16.63.1; Microsoft Corp., Redmond, WA, USA). Graphs were generated using Prism 9 for macOS (GraphPad Software Inc., San Diego, CA, USA). Net and Kaplan-Meier computed survivals at 12-month intervals from diagnosis to 72 months are presented (with CIs). We selected the Kaplan-Meier survival rates and plotted to 5 years after diagnosis, with respect to various tumour features and year of diagnosis. Trends in 2-and 3-year overall survival were plotted for each cancer relative to 2013 outcomes. Survival outcomes were compared using a two-way ANOVA. Patterns of disease were compared using chi-squared tests or t-tests, depending upon the variable. All tests were two-sided, with P < 0.05 taken as statistically significant. Statistical comparisons were conducted in RStudio Version 1.6.0 or Prism 9.

Ethics Approval, Consent and Data Availability
This is publicly available, service-collected, non-identifiable data. Ethics Committee approval and consent are not necessary. The full dataset is available to download at: https:// www.cancerdata.nhs.uk/getdataout/bladder.

Stage and Grade at Diagnosis
The stage at diagnosis was broadly similar over time (incidence rates per 100 000 shown in Fig. 2

Treatment
Treatments received in isolation or combination were detailed according to disease stage, patient factors and Cancer Alliance for BC. Distributions varied with stage ( Fig. 3 and Table S2) and social deprivation (Fig. 4) and between Cancer Alliances (Figs. S3-S5).

Urethral Cancer Treatment
No treatments for urethral cancer were recorded.

Survival Outcomes
Overall survival was determined for 98 031 patients with BC and 16 018 with UTUC, treated from 2013 to 2018. Comparisons concerning the year of diagnosis showed no change in survival rates, whether for all BCs, only MIBCs or UTUC (Fig. S6). For BC, overall survival differed markedly regarding the presence of muscle invasion (ANOVA P < 0.001, pTa/Tis vs pT1 vs MIBC, Fig. 5A). For NMI, survival rates were similar for Grade 1 and 2 cancer (ANOVA P = 0.9) and worse for Grade 3 cancers (ANOVA P = 0.002) at 60 months. Women had worse outcomes than men when all stages were combined (Fig. 5B, ANOVA P < 0.001). For MIBC, nonurothelial tumours had lower short-term survival rates than urothelial cancer, although outcomes at 5 years were broadly similar ( Fig. 5C). Patients with MIBC had lower survival rates than those with MI UTUC (ANOVA P < 0.001). Comparisons with other common cancers revealed that MIBC had worse survival outcomes than many and that NMIBC had a similar survival to all prostate cancers (Fig. S7). Trend analysis highlighted a lack of improvement for BC. For example, the relative change in overall survival rates was 0.97 percentage points (95% CI 0.97-0.97) and 0.99 (95% CI 0.99-0.99) at 3 years, Fig. S8 for BC, in contrast to other cancers (e.g., increased survival from oesophageal cancer [relative 1.06 percentage points, 95% CI 1.06-1.06] at 2 and 3 years).

Discussion
The NCRAS data repository provides important information on incidence, prevalence, routes to diagnosis, staging and treatment for BC, UTUC and urethral cancers. The repository includes NMI disease (which had previously been excluded), and improvements have reduced the amount of missing data (e.g., stage missing in only 7% for BC and 21% for UTUC, compared to >30% previously [22][23][24]). This more complete dataset delivers a different picture of BC from that reported before. For example, there is a much higher incidence of BC  (average annual case rate was >18 000 in 2013-2019 compared to 10 000 previously [22]), >150 000 persons are living with a prior diagnosis of BC (Fig. S1), around onethird of NMI patients have lamina propria invasion (pT1). There are wide variations between different Cancer Alliances.
The route to diagnosis findings question current NHS pathways. In particular, most BCs and UTUCs were not diagnosed via 2ww referrals (in contrast to other common cancers, e.g., 48% for head and neck and 48% for prostate cancers were referred as 2ww priorities, Fig. S9). More BC/ UTUCs were referred from their GP as non-2ww cases (41.3%), so it will be essential to understand which symptoms were responsible for this. Haematuria outcome studies report low rates of BC in persons with non-visible haematuria. That risk varies dramatically between different populations with similar symptoms (e.g., men vs women, smokers vs nonsmokers) [25]. Patient surveys reveal affected individuals often complain of irritative bladder symptoms (dysuria, frequency) rather than or prior to developing haematuria [26]. Changes in routine blood tests can be seen 6-9 months before a diagnosis of BC and kidney cancer [27]. The 2wwreferral criteria should be adjusted to include patient features (sex, smoking history, other risk factors) and symptoms (irritative LUTS), as NHS resources and attention are focused on these pathways. BCs diagnosed via the 2ww route have better outcomes (70% overall survival at 36 months) than those via routine GP referrals (60% at 36 months) [https:// www.cancerdata.nhs.uk/routestodiagnosis/survival], and patients with haematuria face fewer delays than those with other symptoms [28]. Another key observation is that around one in 10 patients were diagnosed at an emergency admission. Presumably, these patients had undiagnosed symptomatic or non-symptomatic cancers in the community, and this emergency represents a failure to find these cancers at this earlier stage. Consequently, there is a direct relationship between rising stage and rates of emergency diagnosis, and wide variations are seen between Clinical Commissioning Groups (Fig. S2C). It will be important to learn from Clinical Commissioning Groups that perform better than others and undertake quality improvement initiatives [29]. Reassuringly, few patients were diagnosed through their death certification (Fig. S10), suggesting a high symptomatic presentation rate and a lack of latent disease.
In general, treatment patterns reflected National Institute for Health and Care Excellence (NICE) and European Association of Urology (EAU) guidance [11,30]. For example, for NMIBC, there appeared to have been a switch to include chemotherapy with TURBT [31]; for MIBC the rates of radiotherapy-only were falling (perhaps with broader use of chemoradiotherapy [32]) and more participants undergoing radical cystectomy also received chemotherapy [11]. These findings are similar to those from Cancer Research UK for 2013-2014 for radical cystectomy (23.3% vs weighted average 22.7% for Stage 2-4 BC) and higher than those for radiotherapy (40.1% vs 28.15% and 36.6% vs 28.4% for Stage 2-4 BC, respectively) [https://www.cancerresearchuk.org/ health-professional/cancer-statistics/statistics-by-cancer-type/ bladder-cancer/diagnosis-and-treatment#heading-One]. That many patients with MIBC did not receive radical treatment is disappointing and matches a more detailed report for nonmetastatic MIBC [20]. Whilst associating treatments with patient factors for MIBC is not possible, the population data suggest radical treatment was less likely to be used with increasing age and comorbidity, with unknown ethnicity, with female sex and with higher social deprivation (Fig. 4). There were marked differences in treatment patters between Cancer Alliances, and (as reported previously) women had worse survival outcomes than men [33].
Various limitations require discussion. First, whilst population-wide data deliver large sample sizes for comparisons, they are limited in terms of granularity. Individual patient factors are unknown, observations need further work to understand (e.g., the reason for referral outside the 2ww), and there can be limited levels of detail or granularity (e.g., differences in NIMBC/MIBC rates between regions). This is demonstrated in the treatment data, for which some details are unclear, and there are no data for pTa/pTis cancers. Radiotherapy allocations are often unknown/other [34] and there is no record of immunotherapy. This is particularly impactful in the analysis of NMIBC, in which intravesical BCG is an important treatment. BCG is often not recorded on the hospital Trusts' e-prescribing system because a urologist, not an oncologist, administers it. The collection of BCG treatment is out of scope in SACT, version 3.0.
Importantly it is not possible to match treatments and patient factors with outcomes to understand the relationships between these factors. Second, despite added work, data remain missing, and areas of high missingness may be those where work is needed most urgently (either in treatment pathways or geographic providers). Regardless of these limitations, population-wide data offer a powerful resource superior to self-submitted data [15]. Whilst the latter were useful for comparative outcomes and the design of clinical trials [35,36], population-wide data allow comparisons between diseases and offer the power to identify problems with disease pathways.
A more complete picture of BC and UTUC offers insights into solutions that could improve outcomes. The higher than previously reported incidence suggests early detection programmes could be possible and effective/cost-effective if implemented among populations with a high rate of undiagnosed disease. This is clearly present in some populations of BC [37][38][39]. At the same time, the high prevalence of low-grade NMIBC (which has a low risk of BC death [40]) highlights the importance of estimating the benefit-harm ratio for BC screening programmes before their implementation without overwhelming diagnostic services. Detecting disease early may decrease the load on both the emergency and the 2ww pathways (which require substantially more NHS resources [41]) and lower late-stage presentation, health system resource use, and costs.

Conclusions
The NCRAS dataset offers key insights into contemporary BC, UTUC and urethral cancer within England. Whilst the dataset does not allow matching for individual outcomes or patient features and key parameters can be missing or incomplete, the findings of this study allow us to question the referral pathways, highlighting that one in 10 patients present as emergencies, which suggests that the diagnostic pathway (and 2ww criteria) for BC needs reviewing. Improvements in survival might be delivered through greater use of radical treatment.

Funding
This study was supported by the Yorkshire Cancer Research UK (Grant number RA/2019/R1/004). James W.F. Catto is funded by a UK National Institute for Health Research (NIHR) Research Professorship.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Appendix S1. Dictionary of source data used in this report.